CS7643 Deep Learning - Module 2 (Convolutional Neural Networks)

Lesson 5: Convolution and Pooling Layers

Convolution Layers

Limitation of Linear Layers

  • Having fully connected layers all the way through isn't always the best choice.
  • When input image has lots of layers, this could lead to overparameterization.
    • Ex. 1024 (M) x 1024 (N) pixel image = M*N + bias = hundres of millions of parameter for just one layer.
  • Overparameterization == overfitting and more data needed.

Locality of Features

Lesson 6: Data Wrangling

In [ ]:
 

Lesson 7: Visualization

Lesson 8: PyTorch and Scalable Training

Lesson 9: Advanced Computer Vision Architectures

Lesson 10: Bias and Fairness